Some Design Issues FacingAmerican Bridge Constructors

T. Y. LinChairman of the BoardT. Y. Lin InternationalSan Francisco, California

Charles RedfieldVice President

T. Y. Lin InternationalSan Francisco, California

T his paper attempts to pinpoint somedesign issues which directly affect

bridge constructors. While the discus-sion is aimed primarily at constructorsin North America, many of the ideascan be applied effectively in other lo-cations throughout the world. In par-ticular, five major items will be dis-cussed:

1. Non-prestressed reinforcement— Discusses the necessity and desira-bility of using non-prestressed rein-forcement to control structural behaviorand increase member strength duringvarious construction stages of a seg-mental concrete bridge.

2. Deflection control — Describesthe roles that hinge location and addi-

NOTE: This paper is based on a presentationgiven at the Segmental Concrete Bridge Confer-ence in Kansas City, Missouri, March 9-10, 1982,The Conference was sponsored by the AssociatedReinforcing Bar Producers — CRS1, FederalHighway Administration, Portland Cement Associ-ation, Post-Tensioning Institute, and PrestressedConcrete Institute.

tion of supplementary tendons play incontrolling deflection in segmentalbridges.

3. Choice of deck sections — Dis-cusses the various sections available forsegmental construction, includingsingle-cell and multiple-cell boxes, T's,I's, waffle, and wing sections.

4. Fabrication and erection tech-niques—Compares match casting ofsegments using epoxy joints with con-ventional casting methods includingthe use of open joints and wet joints.

5. Span-by-span casting versus pre-casting — Compares casting segmentson launching trusses and movableforms using span-by-span constructionas distinct from the use of individualmatch cast precast segments. Whole-span or half-span precast elements willbe described. However, the issue ofdouble cantilever construction, i.e.,comparing cast-in-place concreting toprecasting techniques, will not be dis-cussed.

58

The authors discuss five design areas facingprestressed concrete bridge constructors, namely,selection of non-prestressed reinforcement,deflection control, choice of deck sections,fabrication and erection techniques, andspan-by-span casting versus precasting.

Non-prestressedReinforcement

Almost all segmental bridges arepost-tensioned no matter whether theconcrete segments are cast in place orprecast. Moreover, the segments invari-ably incorporate non-prestressed rein-forcement in one form or another.Therefore, practically speaking, almostall segmental bridges are in effect par-tially prestressed even though no ten-sile stress is theoretically permitted.

The computation of stresses for pur-poses of comparison to allowable codevalues has generally been limited toglobal longitudinal stresses. Although.bridge deck stresses are computed inthe transverse direction, some basicsources of stresses, such as local shrink-age and temperature, are not usuallyconsidered, In other words, most codesdeal with global stresses but not withlocalized stresses.

Some localized stresses for whichthere are no accepted methods of cal-culation and no definite allowablestress values are:

(a) Stresses at tendon anchorages.(b) Stresses along curved tendons.(c) Stresses resulting in slab delami-

nation.(d) Stresses at blisters and dia-

phragms.Similarly, the causes of certain local

stresses are often not specified; neitherare the magnitude of such stresses, asfor example:

(a) Local temperature differentialbetween the outside and inside ofslabs, top and bottom of slabs, andaround corners.

(b) Local shrinkage due to non-homogeneous concreting proce-dures.

(c) Local creep due to high localizedstresses.

(d) Stresses resulting from construc-tion procedures, constructionjoints, stage stressing, etc.

(e) Effects of local reinforcing barsand tendons in reducing concreteareas and increasing local stres-ses.

On the one hand, it is very difficult topredict the factors causing thesestresses. On the other hand, even if thestresses could be predicted, the methodof computation or the mathematicalmodel would be too difficult to specifyor formulate. Furthermore, the use ofthe so-called no-tension criterioncannot be applied since most localstresses will be heavily tensile, result-ing in Iocal cracking.

Therefore, in the final analysis, theintroduction of non-prestressed rein-forcement is often the only effectivelysimple solution. However, determiningwhen such reinforcement is required

PCI JOURNAL/July-August 1982 59

I a. EXTERIOR AND INTERIOR FACES OF SLABS

IC. SUNNY AND SHADY SIDES

lb. TOP AND BOTTOM SLABS

li l ili l i l il l l l ll l l I IIIlilili^il'Illllil'lilll!

ICI lil^l1111111i1111^11III III;I'1'I l l'f'1 t i I'lilll

. I I I

II I.I, II^lillI^I,II

I''II;I I

d. RAIN OR SHINE

yr

I e. SUMMER AND WINTER

Fig. 1. Various causes of temperature stresses in a segmental box.

and finding how much is needed tocontrol the structural behavior and toincrease the member strength locallyare difficult questions to answer.

One important local problem thatfrequently arises is temperature stress.Such stresses produced in segmentalboxes can be attributed to several fac-tors:

(a) The exterior and interior faces ofa slab may possess different tem-peratures (Fig, la).

(b) The top and bottom slabs aresubjected to different tempera-tures (Fig. Ib).

(c) The sunny and shady sides of abox obviously would have differ-ing temperatures (fig. lc).

(d) Rain or sunshine will cause dif-

fering effects on the slabs of a box(Fig. Id).

(e) Summer and winter will producevariable temperature differentials(Fig. le).

A difficult question to answer iswhich of the above conditions shouldbe considered in the analysis of a givenbridge. For example, if all of the abovefactors were computed, it is almostcertain that high tensile stresses wouldoccur somewhere in the box undersome combinations.

Can we afford to make that muchanalysis? If so, what would be the al-lowable stress values under variouscombinations?

Obviously, the only reasonable solu-tion would he to provide temperature

60

Fig. 2. An anchorage for tendons terminated along a segment is commonly known as ablister (1-205 Columbia River Bridge).

steel in an empirical manner. A smallamount of temperature reinforcementcan go a long way in controllingstresses in segmental bridges. Butexactly how much steel is needed, noone knows as yet.

Some other examples of causes oflocal stresses are:

(a) Blisters and anchorages (Fig 2).(b) Slabs heavily separated by ten-

dons running in two directionsand closely spaced together (Fig.3).

(c) Stresses during handling, e. g., atwo-cell box during transportation(Fig. 4).

(d) Stress along sharp curves, pro-ducing high localized stresses, re-sulting in failure (Fig. 5).

Deflection ControlOF particular interest to constructors

is the long-term deflection of segmentalconcrete bridge spans. Some doublecantilever bridges in Europe haveshown an excessive deflection at themidspan hinges and as a result thismethod of construction has been dis-couraged in some countries. On theother hand, there are many existingbridges whose midspan hinges havedeflected within acceptable limits andtoday exhibit a smooth riding surface.

Recently, some bridges in both NorthAmerica and Europe have been de-signed with permanent hinges near thequarter-span point instead of at mid-span. In double cantilever construction,this requires a constructional continuity

PCI JOURNAL/July-August 1982 61

}-- fi/ les. :

^ ^r i r6i S,:L1wmE^

J P R 111 _^ ^ tiwnl ^^I

+ r

-'y r =^ rvti^ r n Z

•

^^1^1+ s rrl

_ _

L'

•mss i.^^^•

• L ` wer.:, :.^LR►^;[ MlrwN ^1 Ml. sr•

Rig. 3. Deck slab with closely spaced tendons in two directions. Note that congestedreinforcement will cause difficult concreting.

Fig. 4. A two-cell box segment during handling, requiring diagonal tendons to transferweight of inner web to the outside webs (1-205 Columbia River Bridge).

62

Fig. 5. Curved tendons forcing themselves out of the websof a segmental box (Las Lomas Bridge, California).

through the quarter-span point in orderfor the two cantilevers to reachmidspan. After closure, the midspanpoint is made continuous and thequarter-span hinge is transformed into apermanent hinge. While such a methodhas been and can be successfully car-ried out, it is a costly and fairly complexoperation requiring additional con-struction time.

It is believed that the midspan per-manent hinge can be constructed with-out undue deflection under certainconditions:

(a) Long-term deflection at a mid-span hinge can be more reliablypredicted (including the effect ofcreep, shrinkage, and tempera-

ture differential) for short spansand for spans with high depth-span ratios and uniform sectionsthan is the case for slender, long,and haunched spans. It followsthat if the elastic deflection issmall, then the time-dependentdeflection will also be small.

(h) Midspan deflection can be bettercontrolled if longitudinal pre-stress is used over and above thatrequired by the code. This willresult in more uniform stressblocks at critical points (for exam-ple, near the piers) so that anyvariation in the modulus of elas-ticity, creep factor, friction ex-pectation, weight, and other vari-

PCI JOURNALJJuly-August 1982 63

Fig. 6. Chung Hsiao Bridge, Taiwan. Built of I-girders, with lengths of 165 and 99 ft(50.3 and 30.2 m) span and 264-ft (70.5 m) centers between piers.

ations will not cause serious de-flections at midspan. The additionof these tendons can be less ex-pensive than the constructionprocess needed to provide quar-ter-span hinges, especially inshorter spans.

If deflection control is needed, ductscan be provided to insert additionaltendons:

(c) The addition of unbonded ten-dons across the midspan hinge toraise the midspan point can beused to control the amount of de-flection at midspan and yet permithorizontal movement at the hinge.This technique is very effectiveand has been used in the Rio Col-orado Bridge in Costa Rica.

The above techniques are alterna-tives offered in controlling deflection atmidspan hinges, thus eliminatingquarter-span hinges, which generallyare objectionable from a constructor'sviewpoint. Whether these alternativesare a more economical solution will, ofcourse, depend upon the circum-stances.

Choice of Deck SectionsIn segmental construction there are

many types of deck sections that can beconsidered. So far, the most populartype has been the box shape, particu-larly the single box, which is conve-nient for both precast segments andin-place construction with travelers.The box section can be used up toabout 70-ft (21.4 m) wide roadwaydecks, using transverse post-tensioning.For wider roadways, double-cell ortriple-cell boxes can be utilized, al-though they do require more compli-cated formwork and casting methods.Also, special handling methods areoften needed to transport and erect adouble-cell or triple-cell precast box.

It is important to note that the boxsection is not the only structural solu-tion, and certainly not for short spans.For spans under 140 to 160 ft (42.7 to48.8 in), a box is often uneconomicaland not structurally efficient becausethe bottom soffit serves little structuralpurpose for such spans. Therefore, thesoffit can be eliminated resulting insimpler formwork in addition to re-

64

Fig. 7. Kwang Fu Bridge, Taiwan. Constructed with I-girders placed along four panels at125 ft (38.1 m) forming a 500-ft (153 m) cable-stayed segmental bridge.

ducing the amount of concrete, rein-forcing bars and tendons.

For positive moment resistance, theT-shape is much more efficient than thebox shape. It is only when the span getslong, and the negative moment requiresa heavier bottom soffit near the piers,that a box section becomes more suit-able. Indeed, when the depth at thepiers can be increased, the T-sectionwill be more economical for evenlonger spans.

The I-section, although seldom usedfor transverse segmental construction inthe manner that a box segment is, hasnevertheless been used in longer seg-mented spans with several of themplaced in parallel to form the width ofthe bridge deck. In addition, standard-ized I-beams have been used very ef-fectively as approach spans in combi-nation with the main segmental spans,as was done recently for the HoustonShip Channel Bridge.

I-girders, similar to the AASHTO I-sections, have been economically usedfor spans up to 264 ft (80.5 m) by incor-porating Y-shaped piers, which shorten

the center span to 165 ft (50.3 m), as inthe Chung Hsiao Bridge (Fig. 6). 1-segments have also been used forcable-stayed segmental construction,with four 125-ft (38.1 m) panels forminga span of 500 ft (153 m), as is the casewith the Kwang Fu Bridge (Fig. 7).

The waffle section is another in-teresting type. It uses precast bathtubsplaced upside down, which are joinedto form a bridge deck without epoxy.Concrete is cast between the waffles onall four sides, thus integrating the waf-fles into one deck. Tendons are placedbetween the waffles and post-tensionedin two directions to transfer the load asdesired.

This type of segment will accommo-date wide bridges, which essentiallyspan in two directions, as for examplethe Hegenberger Bridge (Fig. 8). Thewidth of this bridge is over 100 ft (30.5m), with a span of 300 ft (91.5 m) sup-ported at the four tips of Y-piers. Thisdesign reduced the actual suspendedspan to only 160 ft (48.8 m). Note thatwaffle construction is being applied atthe San Juan Airport in Puerto Rico (see

PCI JOURNAL/July-August 1982 65

Fig. 8. Hegenberger Bridge, Oakland, California. Built of precast waffle segmentspost-tensioned together in two directions.

Fig. 9. Airport Bridge, San Juan, Puerto Rico. Constructed with waffle segments.

66

PT /\I

`r ^ -

Fig. 10a. San Francisco Airport Elevated Roadway Bridge. Built of precast wingsegments without match casting or epoxy joinery.

Fig. 9) and has been found to be ex-tremely economical in the accommoda-tion of curved sections. The system fitsparticularly well when columns form amore or less square pattern.

A further interesting section is thewing type element which was initiallydeveloped for the San Francisco Airport(Figs. 10a and 10b). Here the wingswere precast at a conventional long-linepretensioning plant. These precastsegments require only straight preten-sioned tendons, which run along thetop of the wing ribs and resist thetransverse cantilever moments. Thewing segments are placed side by sideleaving a small gap between. They formthe shell, on top of which the deck con-crete is cast. The spinal beams carry thelongitudinal moment and are post-tensioned to transmit the load to thepier supports. This design allows for anextremely economical production and

erection procedure, without the use ofmatch casting or epoxy. The wing ele-ments are spaced in. (12.7 mm) apartand the joints are taped prior to in-situcasting. The pieces are not in directcontact and therefore the question oftolerances is not a problem.

These wing sections were alsoemployed in Bogota, Colombia (Fig.11), where an international biddingcontest showed this design to be themost economical; to-date, six intersec-tions have been built by this method.The wings are supported on steel formsfor the deck and spinal beam concret-ing. These forms span only 16.5 ft (5.0m), thus requiring supports at 16.5-ft(5.0 m) centers and permitting crosstraffic to go under. However, the formsare adaptable and can be lengthened to30 ft (9.2 m) or more if required.

Currently under design is anotherbridge in Vancouver, Canada, where

PCI JOURNAL/July-August 1982 67

Fig. 1 Ob. Finished view of San Francisco Airport Elevated Roadway Bridge.

these wing sections will span up to 138ft (42 m). This type of construction isdefinitely more economical for shorterspans; however, it does consume morein-place concrete and is heavier thannormally precast sections. Neverthe-less, the elements can be produced byconventional pretensioning plants inNorth America and can be erected byconstructors without much experience,

Fabrication and ErectionTechniques

Although epoxy joints have beenused extensively in North America,they frequently create problems for theinexperienced constructor. Also, matchcasting allows little room for adjustmentif dimensional inaccuracies occur dur-ing placement or storage.

It is suggested that other means of

joinery should be explored. For shortspans, some box segments have beenmatch cast but joined without epoxy.For larger spans, wet joints should beexplored, e. g., using formed joints sev-eral inches thick which can be groutedin place. Such joints require specialforming at the connection, but theformwork can be highly mechanized.Furthermore, the formwork may extendover several segments so that forces canbe transmitted across the wet jointsthrough the steel forms, until the grouthas hardened. Then the steel form canbe moved ahead to the next joints.

This method, if explored properly,can he an economical approach, avoid-ing the use of match casting and epoxyjointing altogether. Such a system couldbe developed in North America but willneed close cooperation between designengineer and contractor.

68

VA

Fig. 11. An overcrossing in Bogota, Colombia. Constructedwith precast wing segments.

Span-by-Span CastingVersus Precasting

While double cantilever constructionhas developed essentially along twomethods of construction, namely, (1)match precasting, and (2) travelingforms with in-place concreting, span-by-span construction has barely begunin North America and much work re-mains to be done in this area.

The use of travelers has indicatedthat there is an important place for in-place concreting, due to the availabilityof ready-mixed concrete trucks, mov-able concreting plants, and pumped con-crete. For double cantilever bridges,in-place concreting has proven to beeconomical under certain conditions,specifically for very long spans, curved

soffits, and a relatively small number ofspan repetitions.

In span-by-span construction, the useof launching trusses for in-place con-creting may have a promising future inthe United States. The problems con-cerning launching truss forms travelingfrom pier to pier can be solved byAmerican constructors and engineers.Such trusses can move in a variety ofways. They can be underslung, traveI-ing upon brackets established at piertops. Or they can be overslung, travel-ing above the bridge deck to cast thesegments. The forms can then movealong, leaving the cast elements to helowered into position. Such methodscan be justified when there is enoughrepetition of spans.

It might be mentioned here that in

PCI JOURNAL/July-August 1982 69

Fig. 12. Precast whole spans (in Italy) pushed along completed portion of bridge towardits destination.

Fig. 13. Bridge in Italy. Segments transported on an underslung launching steel truss.

70

some European countries whole precastspans are hauled overland with trussesthat are either overslung or underslung,as indicated by the photographs fromItaly (see Figs. 12 and 13). Note that forheavy or wide bridges, they can belaunched in half-spans.

From the experience gained so far, itappears likely that segmental concretebridges will continue to be constructedusing a variety of methods. There will,therefore, be an important role for bothprecast and cast-in-place construction,either done separately or in combina-tion.

CONCLUDING REMARKSIt appears obvious that in a large

continent, such as North America, thedevelopment of bridge constructiondoes not necessarily have to followEuropean or any other practice. Ofcourse, we can always learn fromothers, but to adapt other methods toconditions here may require an openmind and different approach. Some ofthe peculiarities of American conditionsare enumerated as follows:

1. Since the conditions in NorthAmerica are so diverse, it is not easy todevelop a highly specialized and spe-cific method of construction which willbe established and practiced among allcontractors throughout the continent.

2. Again, because of its large size,conditions are quite different across thecountry. Geographically, labor supplyand a contractor's experience can varygreatly, requiring different constructionapproaches.

3. There are many innovative con-tractors in North America in addition toa large number of contractors who arevery conservative and only accustomedto traditional methods. in order toutilize the advantages of both factions,the design must he made in such amanner that most, if not all, contractorscan combine their inventiveness.

4. There are both large and small

contractors in North America. For cer-tain bridges we should utilize small,local contractors while for others, big-ger contractors must be brought in fromelsewhere. This is unlike Europe,where countries are small and thelarger contractors can cover the entirecountry, perhaps without much com-petition.

5. American contractors are ex-tremely innovative even though theyoften do not work with engineers in de-sign. North America has an establishedindustrial base in machinery designs andsupply. If given the opportunity,mechanized production for multiplespans can become extremely economi-cal.

6. In America there are well over500 precasting plants. Although theirequipment, experience, and capabil-ities are somewhat different, mostplants have product versatility and in-novative personnel.

7. There are several experiencedpost-tensioners who innovate and com-pete. They can supply tendon hardwareor post-tensioning technology asneeded. It should be possible to com-bine their expertise with the construc-tors in an optimum manner.

Finally, while we are interested inintroducing segmental constructionmethods from other countries, weshould also broaden our own view-point. We should visualize segmentalconstruction as consisting of differenttypes and forms of segments, differentmethods of precasting or concreting,and different ways of transportation,erection and joinery.

From that viewpoint, segmental con-crete bridge construction is only begin-ning in North America. Unquestionably,the practice will expand and vary as ex-perience is gained and as the engineerswork together with the constructors tocome up with economical ways of de-signing and building bridges which inturn will also attract the attention of theinternational engineering profession.